From 990af8b9b8dab3bcfde3f93e21d89b3a2f41dcd8 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Tue, 4 Aug 2020 10:38:10 -0700 Subject: more update keys and cases for chocula importer --- python/fatcat_tools/importers/chocula.py | 16 +++++++++++----- 1 file changed, 11 insertions(+), 5 deletions(-) (limited to 'python/fatcat_tools/importers/chocula.py') diff --git a/python/fatcat_tools/importers/chocula.py b/python/fatcat_tools/importers/chocula.py index 1683c500..356ffe42 100644 --- a/python/fatcat_tools/importers/chocula.py +++ b/python/fatcat_tools/importers/chocula.py @@ -53,7 +53,8 @@ class ChoculaImporter(EntityImporter): extra = dict() for k in ('urls', 'webarchive_urls', 'issne', 'issnp', 'country', - 'sherpa_romeo', 'ezb', 'szczepanski', 'doaj', 'languages'): + 'sherpa_romeo', 'ezb', 'szczepanski', 'doaj', 'languages', + 'ia', 'scielo', 'kbart', 'publisher_type', 'platform'): if row['extra'].get(k): extra[k] = row['extra'][k] @@ -115,17 +116,21 @@ class ChoculaImporter(EntityImporter): do_update = True if set(ce.extra.get('webarchive_urls', [])) != set(existing.extra.get('webarchive_urls', [])): do_update = True - for k in ('ezb', 'szczepanski', 'doaj'): + for k in ('ezb', 'szczepanski', 'doaj', 'publisher_type', 'platform'): if ce.extra.get(k) and not existing.extra.get(k): do_update = True + for k in ('kbart', 'ia', 'doaj'): + # always update with these fields + if ce.extra.get(k) and ce.extra[k] != existing.extra.get(k): + do_update = True if ce.publisher and not existing.publisher: do_update = True if ce.wikidata_qid and not existing.wikidata_qid: do_update = True if do_update: - existing.wikidata_qid = ce.wikidata_qid - existing.publisher = ce.publisher + existing.wikidata_qid = existing.wikidata_qid or ce.wikidata_qid + existing.publisher = existing.publisher or ce.publisher existing.container_type = existing.container_type or ce.container_type for k in ('urls', 'webarchive_urls'): # update, which might clobber, but won't remove @@ -135,7 +140,8 @@ class ChoculaImporter(EntityImporter): # all URLs found to be bad), but being conservative for now so # we don't clobber human edits for k in ('issne', 'issnp', 'country', 'sherpa_romeo', 'ezb', - 'szczepanski', 'doaj'): + 'szczepanski', 'doaj', 'ia', 'scielo', 'kbart', + 'publisher_type', 'platform'): # update/overwrite, but don't remove any existing value if ce.extra.get(k): existing.extra[k] = ce.extra[k] -- cgit v1.2.3